AgentSkillsCN

Pagerduty Service Reader

读取一篇 PagerDuty 服务,提取近期事件的相关详情,返回约 500 个 Token 的摘要。

SKILL.md
--- frontmatter
description: "Read one PagerDuty service and extract details with recent incidents. Returns ~500 token summary."
allowed-tools: ["Bash"]
model: haiku
context: fork
agent: Explore

PagerDuty Service Reader

You read ONE PagerDuty service and extract details including recent incidents (~500 tokens).

Input Format

The user provides: {service_id} | {research context}

Examples:

  • PG7CZUT | understanding ownership and on-call
  • PABC123 | recent incident patterns for this service

Instructions

  1. Extract the service_id from the input (ID before the |)
  2. Run the pagerduty-service-reader script:
bash
~/.dataops-assistant/bin/pagerduty-service-reader.sh {service_id}
  1. Read the research context - it tells you WHAT to extract
  2. Extract ONLY information relevant to that context from the script output
  3. Return structured output with relevance rating

Key principle: You are NOT summarizing the whole service. You extract what matters for THIS research question.

ID Format Detection

The script automatically detects ID format mismatches:

  • Service IDs: Typically shorter, 7 characters, start with P (e.g., PG7CZUT)
  • Incident IDs: Typically longer, often start with Q (e.g., Q0RIJJZL24RC6W)

If given an incident ID, the script will error and suggest using pagerduty-incident-reader instead.

Output Format

code
SERVICE: {name}
ID: {service_id}
STATUS: {active|disabled|maintenance}
DESCRIPTION: {description or "none"}

RESEARCH CONTEXT: {echo what we were looking for}

TEAM: {team_name} ({team_id})
ESCALATION POLICY: {policy_name}

ON-CALL (current):
- L1: {name} until {date}
- L2: {name} until {date}

RECENT INCIDENTS ({count} in last 30 days):
| # | Status | Urgency | Title | Created |
|---|--------|---------|-------|---------|
| {num} | {status} | {urgency} | {title} | {date} |

INCIDENT PATTERNS (if relevant):
- {Pattern observation relevant to research}

RELEVANCE: {high|medium|low} - {brief explanation}

Rules

  • MAX ~500 tokens output
  • Extract only what's relevant to research context
  • If service has minimal relevance, say so and keep output brief
  • List MAX 5 recent incidents in table
  • Note patterns only if relevant to research (e.g., "3 similar alerts this week")
  • Include RELEVANCE rating